Traditionally, enterprises have relied on increases in capital investment and in people for economic growth and progress. With globalization, these same enterprises, now faced with stiff competition and strict budgets, have turned their attention to Artificial Intelligence (AI)--a forefront investment strategy for growth and progress continuity.
With AI, robots and intelligent machines can take the form of physical capital. And unlike conventional capital such as machines and buildings, AI can improve growth over time due to its self-learning capabilities.
Using AI as a new factor for production, enterprises can expect significant growth opportunities. For example, AI can help in predicting production failure points, enabling goods on manufacturing belts to take alternate routes. Anomaly detection and corrective action such as this minimizes losses and increases revenue. By predicting the failure of oil drilling machines, airplane engines, and any machine failure, AI provides predictive maintenance, reducing not only loss in revenue but the loss of lives.
AI’s predictive maintenance also includes the identification of supply and demand levels in many industries such as pharmaceuticals, insurance, and banking. For example, AI can predict pharmacy inventory levels by zip code so that pharmacies can stock up on pharma material and correctly minimize loss of revenue while increasing customer satisfaction. AI can predict the risk associated with underwriting commercial or consumer insurance for carriers and help insurance companies write better insurance policies. Banks using AI experience better cross-sell and upsell opportunities.
AI: Sense ⇒ Analyze ⇒ Predict/Recommend/Acted
- Sense - AI senses the correct data needed, collects that data for the right information, and lastly, collates and prepares the data for analysis.
- Analyze - AI either analyzes the prepared data and provides facts about that data, or transforms the data and represents the data in visualizations. Furthermore, when AI analyzes the data, it also can collate the data or create data clusters, and sometimes when the data is not ready for processing, it can extract relevant information using machine learning techniques that make use of Natural Language Processing (NLP); this extracted information can be called on for further processing.
- Predict/Recommend/Act – AI predictions or predictive models exploit patterns found in historical and transactional data to identify risks and opportunities. Models capture relationships among many factors to allow assessment of risk or potential associated with a set of conditions, guiding decision making for candidate transactions.
AI growth is stemmed from three factors – (1) abundant low-cost computing power (cloud or non-cloud), (2) growth in big data (heterogeneous poly-structured data), and (3) a growing need to gain near-real-time insights into data.
AI Intelligent Assistants
Intelligent assistants are a manifestation of AI, leveraging both AI and natural language processing. Due to the many technological media-spins, many have assumed that AI has been “hyped” and any real advancements are far into the future.
As Bradley Voytek said in his blog (graph below is courtesy of Bradley Voytek), “Data Science is simply plummeting toward the hype cycle ‘Trough of Disillusionment.’” But in terms of AI, this is not disillusionment for all industries.
Tesla automobiles, Google automobiles, Google Translation, Trifacta Data Transformation rules, unmanned aerial vehicle (UAV), unmanned aircraft system (UAS), and other companies have already used AI to advance automation, providing benefits to both consumers and non-consumers.
Although AI adoption is still slow in many IT verticals, those who are skeptics or are disillusioned should think now and plan for adopting AI in their solutions. The reason: AI is becoming more user-friendly than traditional UI.
Artificial Intelligence as the New UI of the Future
One can say that the term "Artificial Intelligence (AI)" applies to a machine when it mimics "cognitive" functions that are associated with the human mind such as "learning" and "problem-solving."
AI is moving from back-end processing for specific verticals in the enterprise to more sophisticated roles that apply cognitive functions within technology interfaces. Take, for example, Amazon Eco Alexa, which moved from the Amazon research labs to the homes of consumers. Eco Alexa has no user interface or UX. It uses spoken words and Natural Language Processing (NLP) combined with AI to serve consumers music, news, and security for their homes, etc.
Today more than 3 million consumers are using Alexa without a user interface and without any training. In the case of autonomous driving vehicles, the use of neural networks through computer vision and sensors make these vehicles easy to operate. As a move from dashboard programming of locations from point A to point B, AI provides several key differentiators that Tesla and the like have adopted.
As AI matures, obstacles or hindrances encountered in adopting AI are disappearing; thus, adoption of AI is increasing. In the AI move from research labs to the consumer’s home, AI provides frictionless intelligence with a simple and smart interface. AI has also branched into consumer homes in which products (Amazon), music (Spotify), and movies (Netflix), are recommended through AI learning of consumer behavior and consequently, without the need for consumers to use any user interface.
AI has also made its way into the hospitality arena – controlling room HVAC, recommending services to customers, etc., and again without the need for a user interface. In this area in which AI orchestrates, it has become highly-sophisticated in accomplishing tasks, offering collaboration across experiences and channels (often behind-the-scenes). AI not only curates and acts based on its experiences but also learns from interactions to help suggest and complete new tasks.
For IT and CIOs, AI’s current capabilities are a good thing, as it moves past the hype and sets the ground work for carving a niche, laying the foundations for a new field to reap AI benefits. So, what kind of tools or architecture should IT and CIOs be looking at as they take the leap forward to AI? Choosing the right architecture for many IT departments is a cost-conscious choice, in which decisions are based on cost-effectiveness and budgetary constraints.
What is IBDaaS and how it paves the path to AI?
Traditionally, the following service models exist: IaaS (Infrastructure as a Service), PaaS (Platform as a Service) and SaaS (Software as a service):
- IaaS - Examples of IaaS are virtual machines, networks, storage, or servers, as the most basic building block. Amazon, Azure and Google provide such IaaS services.
- PaaS – Includes commonly-employed software such as web and database servers, or Hadoop and its ecosystem.
- SaaS – Includes services that are still generic but provides user facing services like web email, content or customer relationship management systems.
But for any of these service models, users must write solutions on top of IaaS, PaaS or SaaS stacks. At the platform level, Hadoop or an alternative distributed compute and storage technology naturally builds the core of a BDaaS (Big Data as a Service). Consequently, any BDaaS solution includes the PaaS layer and potentially SaaS and/or IaaS.
Some vendors focus on Hadoop and an optimized infrastructure for performance, providing offerings for a combined IaaS and PaaS. Other vendors focus on optimizing Hadoop and features for productivity and an exchangeable infrastructure, offering a combined PaaS and SaaS approach. Another important option is a fully vertically-integrated solution with BDaaS that combines the performance and feature benefits of an optimized platform and optimized Big Data.
In this IBDaaS (Integrated Big Data as a Service) stack, besides optimized and fine-tuned PaaS, one desired architecture for SaaS layer consists of the following: 1) a Collector 2) a Storage Layer 3) Connectors Layer 4) an Analyzer and 5) an AI Workbench.
- Collector should have the capability to ingest large volumes of data at high speed from many sources containing poly-structured data from heterogeneous sources.
- Storage Layer should have the capability to access optimized storage from the PaaS layer and be able to combine other federated data stores such as SQL, No-SQL, flat files, and multi-media data. It should provide low maintenance overhead.
- Connectors Layer should have the capability to interact with external systems such as archival storage (Cloud/non-cloud), Directory Services (AD/LADP) and security services, email/SMS/REST services, etc.
- Analyzer should be able abstract and access data from federated data stores, combine the data and provide a consistent view of data from advanced analytics and workflows.
- AI Workbench should provide the capability for self-learning analytics and associated workflows using advanced machine learning (AI) algorithms. The data models built using the AI algorithms should generate actions in near real-time for anomaly detection, recommendations, predictive maintenance, Risk management, Fraud Detection, etc.
As big data increases from day-to-day and more service models are emerging, CIOs are looking to build better scalable solutions with lower budgets and fewer resources. They need a more-cookie cutter approach to building solutions. CIOs need to choose an architectural stack that enables them to build or buy short-term solutions and scale the stack to accommodate building strategic AI solutions.
CIOs do not want to duplicate the data for each solution that they build. They need an architecture that leverages the same data (stored once) and is used for multiple solutions. CIOs do not want to build multiple SaaS stacks to build multiple solutions either. They want to pick architecture stacks that provide them with the ability to build or integrate multiple solutions or integrate with some of the existing solutions.
CIOs are decreasingly willing to take on the complex challenges of building their own data architecture and related SaaS layer, and increasingly want to focus on the value of their existing solutions by adding domain-specific processes. IBDaaS offers an architecture that provides performant, scalable, and customizable solutions. It also provides an architecture that allows CIOs to build new solutions on the same stack.
IBDaaS allows IT to start building “Now” AI-based solutions for the future
Which vendors are offering IBDaaS?
There are only a few vendors who provide Big Data as a Service (BDaaS)– Bluedata and Quebole. But, there are no vendors today offering Integrated Big Data as a Service (IBDaaS). iFusion Analytics from Innominds has taken a new approach in building IBDaaS stack and will provide a revolutionary technology for CIOs.
About iFusion Analytics
iFusion Analytics is a patented scalable and distributed platform that provides AI solutions as Integrated Big Data as a Service (IBDaaS). It comes with “out-of- the-box” rich machine learning algorithms, along with out-of-the-box IIoT (Industrial Internet of Things) and Anomaly detection solutions. iFusion Analytics collects poly-structured data from heterogeneous sources and federated data stores. It cleans data, curates the data, and makes the data ready for data analysts and data scientists to accelerate insights and build solutions at reduced cost.